Research
Our work centers on spatial intelligence, is enabled by agentic intelligence, and advances toward interactive and embodied spatial intelligence, grounded in urban science.
Spatial Intelligence
Grounded in humans, cities, society, and physical environments.
Represent · Benchmark · Reason
Spatial Foundations
Models, benchmarks, and multimodal foundations that establish how machines represent and reason about cities as open spatial systems.
Perceive · Interpret · Ground
Reality-Grounded Understanding
Spatial perception and reasoning grounded in visual evidence, socioeconomic signals, human wellbeing, and the lived realities of cities.
Model · Simulate · Act
World Models & Interactive Intelligence
Generative world models and spatial agents that move beyond passive understanding toward simulation, interaction, and action.
Reason · Learn · Adapt
Agent Foundations & Reasoning
Foundations for agents that reason, coordinate models, learn from feedback, and adapt across tasks and environments.
Perceive · Plan · Act
Embodied & Interactive Agents
Agents that connect perception and reasoning with planning, simulation, and action in dynamic spatial environments.
Collaborate · Simulate · Evolve
Multi-Agent & Social Intelligence
Multi-agent systems for modeling collective behavior, simulating social processes, and studying how agents interact and evolve.
Research Foundations
Earlier Research Archive
A systematic summary of earlier work in individual mobility behavior, citywide spatio-temporal prediction, and operational urban decision systems.